基于改进ORB的图像特征点匹配分析  

Analysis of Image Feature Point Matching Based on Improved ORB

作　　者：陈心如 CHEN Xinru(Liaoning University of Engineering and Technology,Liaoning 125105,China.)

机构地区：[1]辽宁工程技术大学,辽宁125105

出　　处：《电子技术(上海)》2025年第1期102-105,共4页Electronic Technology

摘　　要：阐述针对ORB在特征点匹配中尺度不变性的不足,提出一种基于SIFT(Scale-Invariant Feature Transform)的改进ORB(Oriented FAST and RotatedBRIEF)算法。首先,建立稳定极值点检测的尺度空间,将检测到的稳定极值点作为具有尺度不变性的特征点。其次,利用ORB描述符对特征点进行描述,最终形成具有尺度不变性和旋转不变性的二元描述符。最后,根据汉明距离进行点匹配。实验结果表明,该算法在考虑尺度不变性的情况下具有良好的匹配性能。This paper expounds the minor drawbacks of scale invariance in feature point matching using ORB,and proposes an improved ORB(Oriented FAST and Rotating BRIEF)algorithm based on SIFT(Scale Invariant Feature Transform).Firstly,it establishes a scale space for stable extremum point detection,using the detected stable extremum points as feature points with scale invariance.Secondly,it uses ORB descriptors to describe feature points,ultimately forming a binary descriptor with scale invariance and rotation invariance.Finally,it performs point matching based on Hamming distance.The experimental results show that the algorithm has good matching performance considering scale invariance.

关 键 词：智能算法 图像特征 SIFT 改进ORB 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] TP183[自动化与计算机技术—计算机科学与技术]